Homa Priya Tarigopula
I did my undergrad in India at the IIITDM, Kancheepuram in Electronics and Communication Engineering. It was during this time, I developed an interest in Artificial Intelligence, and went ahead to do a Masters in AI at the University of Edinburgh. I worked on developing Unsupervised Models for Visual Saliency based on the psychological underpinnings of the brain, which funnelled my interests towards brain-inspired AI. I then pursued another Masters in Cognitive Neuroscience at the University of Trento, Centre for Mind/Brain Sciences (CiMEC). At CiMEC, I worked on developing Deep learning models to predict Human Similarity Judgements from behavioural and neural data. During my PhD, I'll be working on using Deep Learning to understand spatial navigation in the mouse brain.
Research interests and hobbies
My primary interests lie in computational neuroscience, deep learning and navigation. I will implement and study the dynamics of transformer networks that are trained to navigate and use the findings to interpret navigation in mouse brains. As for hobbies, I like hiking.
Bio-inspired neural networks for navigation
Transformer networks are a recent advance in Natural Language Processing, which have shown the state of the art performance in autoregressive modelling. In this project, we use these transformer networks to solve a navigation task and study the dynamics of these networks. It has already been found that Recurrent Neural Networks trained for navigation spontaneously form structures that are similar to the grid cell structures found in the brain. Here, we use Transformer networks to understand if they solve navigation similarly, and how these representations can be reused for other tasks.